Visual Binaries in the Orion Nebula Cluster

Bo Reipurth1, Marcelo M. Guimaraes1,2, Michael S. Connelley1,3, and John Bally4

ABSTRACT

We have carried out a major survey for visual binaries towards the Orion Nebula Cluster,using images obtained with the Advanced Camera for Surveys on the Hubble Space Telescopethrough an Hα filter. Among 1051 stars more than 60′′ from θ1 Ori C, we have selected 781stars that fulfill criteria for membership in the Orion Nebula Cluster. Among these, we find 78multiple systems (75 binaries and 3 triples), of which 55 are new discoveries, in the range from0.′′1 to 1.′′5. We perform a statistical study of the 72 binaries and 3 triples that have separationsin the limited range 0.′′15 to 1.′′5, within which we need no incompleteness correction. An analysisof the stellar density in our images suggests that of these binaries, 9 are line-of-sight associations.When corrected for this, we find a binary fraction of 8.8%±1.1% within the limited separationrange from 67.5 to 675 AU (counting the 3 triples as 6 binaries). The field binary fraction in thesame range from Duquennoy & Mayor (1991) is a factor 1.5 higher. Within the range 150 AU to675 AU that overlaps with the study of binaries in T Tauri associations by Reipurth & Zinnecker(1993), we find that the associations have a factor 2.2 more binaries than the Orion NebulaCluster, in approximate agreement with earlier results based on data from the inner Trapeziumregion with small-number statistics. The binary separation distribution function of the OrionNebula Cluster shows unusual structure, with a sudden steep decrease in the number of binariesas the separation increases beyond 0.′′5, corresponding to 225 AU. We have measured the ratio ofbinaries wider than 0.′′5 to binaries closer than 0.′′5 as a function of distance from the Trapezium,and find that this ratio is significantly depressed in the inner region of the Orion Nebula Cluster.The deficit of binaries with larger separations in the central part of the cluster is likely due todissolution or orbital change of the wider binaries during their passage through the potential wellof the inner cluster region. All of our primaries appear to be T Tauri stars with the exception ofone Herbig Ae/Be star, and there are indications that a substantial number of secondaries couldbe brown dwarfs.

Subject headings: binaries: visual — stars: pre-main sequence — stars: low-mass, brown dwarfs — openclusters and associations: individual(Orion Nebula Cluster) — techniques: high angular resolution

1. INTRODUCTION

Since the first detection of pre-main sequencebinaries by Joy & van Biesbroeck (1944) and Her-big (1962), major advances have been gained inour understanding of the formation and early evo-

1Institute for Astronomy, University of Hawaii, 640 N.Aohoku Place, Hilo, HI 96720 (reipurth@ifa.hawaii.edu)

2Departamento de Fisica, UFMG, Caixa Postal 702,30.123-970, Belo Horizonte, MG, Brazil

3NASA Ames Research Center, Moffett Field, CA 940354CASA, University of Colorado, 389 UCB, Boulder, Col-

orado 80309-0389

lution of binaries. One of the key results is thatbinaries are about twice as common in T Tauriassociations as among field stars (e.g., Reipurth &Zinnecker 1993, Simon et al. 1995, Duchene 1999,Ratzka et al. 2005). Studies of embedded sourcessuggest that multiple systems may possibly bemore prevalent among Class I sources than amongthe more evolved T Tauri stars (e.g., Duchene etal. 2004). Dynamical evolution among higher-order multiples is likely to be important during theembedded phase, leading to the occasional ejec-tion of lower-mass members and the formation ofpowerful jets (Reipurth 2000).

1

The Orion Nebula Cluster (ONC) is the nearest(d∼450 pc) star forming cluster, with an age of lessthan 1 Myr, and it has been extensively studiedat multiple wavelengths (e.g., Herbig & Terndrup1986, Hillenbrand 1997, Getman et al. 2005 andreferences therein; for earlier work see the reviewby Herbig 1982). It is well known that the massivestars in the Trapezium have a very high binaryfrequency (e.g., Preibisch et al. 1999). Binarityof the low-mass population of the central part ofthe ONC has been studied extensively, and it hasbeen suggested that the binary frequency is lowerthan for the dispersed low-mass young populationin associations and comparable to the binary fre-quency in the field (Prosser et al. 1994, Padgettet al. 1997, Petr et al. 1998, Simon et al. 1999,Kohler et al. 2006).

In this paper, we report the results of a majorbinary survey of an extensive region centered onthe ONC, based on a large Hα imaging survey us-ing the Hubble Space Telescope. We have detected78 multiple systems, of which 55 are new discover-ies, in an area of 412 arcmin2 that excludes a cir-cular area with radius of 60′′ centered on θ1 Ori C.We discuss the membership of these binaries in theONC, analyze their binary properties, and discussthe binary separation distribution function and bi-nary formation in the ONC.

2. OBSERVATIONS

The present survey is based on our recent studyof the Orion Nebula using the Wide Field Cam-era of the Advanced Camera for Surveys onboardHST. During program GO-9825, we observed 26ACS fields with the F658N (Hα+[NII]) filter andwith an exposure time of 500 sec per pointing. Intotal our Hα survey covers an area of 415 arcmin2

of the central Orion Nebula. Due to the high stel-lar density near the Trapezium we have excludeda circular area with radius of 60′′ centered onθ1 Ori C, so the area investigated is 412 arcmin2.The exquisite resolution of the HST combinedwith the 0.′′05 pixel-size of the ACS offers a uniqueopportunity to detect close binaries. We examinedall stellar sources in the images by eye at a rangeof contrast levels, allowing us to pick up faint aswell as bright companions. Thanks to the very sta-ble point-spread function across the images, thiscould be done in a rather homogeneous fashion.

We discuss the completeness of this procedure inSect. 3.3. We used reduced images that were com-bined through MULTIDRIZZLE. The faintest starwe measured had a magnitude of 21.5 in the F658filter. Pixel coordinates for both primaries andsecondaries were determined with 2-dimensionalGaussian functions, and separations and positionangles calculated from these coordinates. Relativefluxes were determined using aperture photometrywith corrections for the sky background. For fur-ther details of the observations, see Bally et al.(2006).

3. SELECTION OF SAMPLE

3.1. Survey Area

The region of the ONC we have studied isshown in Figure 1, which outlines the mosaic of 26ACS fields. One field had to be slightly turned toacquire a guidestar. We examined all stars (1051)in these fields with the exception of stars withinan exclusion zone of 60′′ around θ1 Ori C.

3.2. Separation Limits

All previous HST studies of binarity in theONC were done in the region around the Trapez-ium, where line-of-sight pairs due to the high stel-lar density is a major issue, and thus focusedmostly on sub-arcsec binaries. In contrast, ourstudy covers a much larger area including the out-skirts of the ONC. We have also excluded the in-ner region with radius of 60′′, so contaminationis much less of an issue. Based on our observedstar density, we can calculate the fraction of ourbinaries that are not physical pairs for differentseparation limits. From this we have chosen anupper separation limit of 1.′′5, which implies an11% contamination, as described below.

3.3. Incompleteness

We have tested our completeness of companionsby blindly “observing” real stars with artificiallyadded companions (scaled from real stars in theimages) at random separations, position angles,and flux ratios. We find that our detections areessentially complete down to separations of 0.′′1,a result made possible by the very sharp pointspread function of the HST/ACS system. Figure 2shows two panels with plots of ∆m vs. separation

2

in arcseconds. The left panel shows the actualobservations of the binaries we detected, while theright panel shows our artificial binaries. Filled cir-cles indicate companions we were able to identify,and empty circles indicate those we failed to see.Our detection limit is essentially complete to 0.′′1except for the largest brightness differences, butvery few binaries are found with large flux dif-ferences (see Sect. 4.5). To be certain that weare complete, we considered binary companions inthe more limited separation range from 0.′′15 to1.′′5. However, a much more severe problem withcompleteness is due to the strong and highly struc-tured emission from the surrounding HII region, aproblem that is compounded by our use of imagestaken through an Hα filter. In some areas, espe-cially near the Trapezium, our ability to detectfaint stars is diminished, an effect that is unquan-tifiable but should be kept in mind.

Most binaries that have been found prior to thisstudy were identified using infrared techniques.Whereas all binaries we have detected have beenchecked against lists of previously discovered bi-naries, we have not made any systematic effort tocheck if all these earlier binaries are also detectablein the optical. From random checks it appearsthat many of these infrared-detected binaries arenot visible in our optical images.

3.4. Membership of the ONC

The main source for membership information inthe ONC are astrometric studies, principally thestudy by Jones & Walker (1988), which covers theinner 20′×30′ of the ONC and thus includes ourarea of study. It is based on photographic I-bandplates, which favors red and reddened stars. Ofthe 1051 stars in our images, 655 stars are listedin the Jones & Walker catalog, and of these 596have membership probabilities higher than 93%.The large majority of our binaries are not resolvedin the JW survey, and thus the proper motionscan be affected by relative brightness variationsof the components that can shift the photocen-ter. We have seen such variability in a binary ly-ing in the overlap region between two fields, andthus observed twice. We have also found starswith clear pre-main sequence characteristics thathave membership probability of 0%, indicatingthat astrometric information can be used to sup-port membership, but should not be used to ex-

clude membership. We are therefore complement-ing the proper motion membership analysis withother sources, such as X-ray emission, Hα emis-sion, and variability. Getman et al. (2005) has de-termined that 1373 stars from the 1616 sources inthe COUP survey, which largely overlaps with ourfield (see Figure 1), are Orion members. Amongour stars, we find 658 which are detected by COUPand also are classified as ONC members. Finally,we use the presence of Hα emission (partly fromthe literature and from our own unpublished deepsurvey), as well as irregular variability as listed inthe General Catalogue of Variable Stars. A totalof 99 stars were found to have Hα emission. Notethat we do not use near-infrared excess as a mem-bership criterion, since our binaries by default arelikely to have infrared excesses due to their mostlylower-mass companions. In total, we find 781 starsthat have at least one (and commonly several) ofthese membership characteristics.

3.5. Binaries and Line-of-Sight Pairs

Among the 781 known ONC members detectedin our HST images, we found 72 binaries and 3triples in the range from 0.′′15 to 1.′′5. These arelisted in Table 1, where each triple is listed astwo binaries, in the manner of counting of Kuiper(1942). The first and second columns list the Jones& Walker ID and the probability of the star to bea member of the ONC based on its proper motion,respectively. The third column lists the Parenagonumber, the fourth column lists the name in theGeneral Catalog of Variable Stars, the fifth col-umn lists the COUP number, and the sixth col-umn lists the number given by Hillenbrand (1997).The right ascension and declination for equinox2000 are listed in the seventh and eighth columns.The ninth column lists the position angle of thesystem and the tenth column lists its separationin arcsec. The apparent magnitude in the I band(from Hillenbrand 1997) is listed in the eleventhcolumn and the difference in magnitudes, in theHα filter, between the primary and secondary islisted in the twelfth column. The position angletowards and separation (in arcsec) to θ1 Ori C arelisted in columns 13 and 14, respectively. Then fol-lows the spectral type as listed by SIMBAD. Thesixteenth column lists the character of member-ship for each binary (P: proper motion; X: COUPONC source; H: Hα emission; V: irregular vari-

3

able). The last column lists the discovery paper ifa binary was known prior to this work.

Additionally, we have found 3 binaries amongthe ONC members outside the 60′′ exclusion zonewith separations between 0.′′11 and 0.′′15. Theyare listed in Table 2. We have furthermore ex-amined the optically visible ONC members withinthe exclusion zone, and have found another 3 bina-ries with separations less than 0.′′4, an (arbitrarilychosen) upper limit we imposed due to the highstellar density in that region. Two of these arenew discoveries. Many more binaries have beenfound by earlier studies of the central region (e.g.,Prosser et al. 1994, Padgett et al. 1997, Petr etal. 1998, Simon et al. 1999, Kohler et al. 2006),but most are only detectable in the infrared, sincemost cluster members are hidden by extinction;these three binaries are the only we could detectin our images in the range 0.′′1 - 0.′′4. We do notinclude the abovementioned binaries in our statis-tical analysis, but list them in Table 2. Finally, wehave searched the 270 stars outside the exclusionzone for which we have no evidence for member-ship, and have found another 7 binaries. Someof these may be foreground or background stars,but others may turn out to be young stars, andwe also list them in Table 2. The columns in Ta-ble 2 are the same as in Table 1, except we haveadded an additional column, which characterizesthe binaries.

All of our binaries with known spectral typesare late-type stars, presumably classical and weak-line T Tauri stars, with the exception of Pare-nago 2149 (JW 945), which appears to be a HerbigAe/Be star.

In Figure 3a,b we show figures of the 78 multi-ple systems we have identified among ONC mem-bers outside the exclusion zone, as well as thethree close binaries (<0.′′4) found inside the ex-clusion zone. Each stamp is 2′′ on a side. Some ofthe binaries are particularly interesting for a vari-ety of reasons, e.g. V1118 Ori is a famous EXor,other binaries or companions are likely substellarobjects, and some are associated with jets or pro-plyds. In the Appendix, we provide more detailsabout these cases.

Could some of the wider binaries in the ONC bedue to contamination by line-of-sight pairs? Thesurface density of stars in the ONC is a steeply de-clining function of distance from the center of the

cluster, with only relatively minor fluctuations dueto subclustering (either real or caused by extinc-tion variations, Figure 4). We have determinedthe stellar density Σ in an area with radius 30′′

around each of the 781 ONC members, and havedetermined the probability P of finding an unre-lated star within a distance θ from each primaryusing the expression P = 1 − e−πθ

2Σ (Correia etal. 2006), where we have set θ to 1.′′5. The prob-ability P is principally a function of the distancefrom θ1 Ori C, with smaller variations due to localinhomogeneities in the cluster density. Figure 5shows the distribution of probability of a line-of-sight association for each of the 781 ONC membersas a function of distance from θ1 Ori C. The fig-ure mostly follows the radial stellar density curvein Figure 4, although the effect of local small-scalegroupings is visible.

The sum of contamination probabilities for the781 stars is 911%. In other words, among the bina-ries detected we are likely to have roughly 9 line-of-sight doubles. In the following we correct forthese false binaries.

4. RESULTS

4.1. Binary Fraction in ONC

The detection of 72 binaries and 3 triples withseparations in the range 0.′′15 - 1.′′5 among the781 ONC members must be corrected for the es-timated 9 binaries that are likely to be due toline-of-sight pairing. Counting the three triplesas six binaries, we then have 69 physical binaries,which implies a 8.8%±1.1% binary fraction bf inthe interval from 67.5 to 675 AU, or a multiplic-ity frequency (where the 3 triples are counted as1 system each) of 8.5%±1.1% (here and in the fol-lowing the error estimates are 1σ and are derivedusing Poisson statistics). Petr et al. (1998) found4 binaries in the separation range 0.′′14 to 0.′′5 inthe inner 40′′×40′′ of the Trapezium, correspond-ing to 5.9%±4.0%. In the same limited range ofseparations we find 50 binaries, corresponding to6.4%±0.9%. The numbers are consistent withintheir errors.

Reipurth & Zinnecker (1993) observed nearbyT Tauri associations and found 38 binaries (of asample of 238 stars) in the range 150-1800 AU.The range we study statistically is 67.5-675 AU.We have derived the number of binaries in the

4

common range 150-675 AU, and find 11.8%±2.2%for the associations and 5.3%±0.8% in the ONC(for simplicity we here count a triple as two bi-naries). Both surveys represent observed binaryfractions, but the Reipurth & Zinnecker survey hasnegligible incompleteness correction in the chosenrange. For our present survey, we have 9 line-of-sight pairs among our 78 binaries, and when weanalyze their statistical distribution for differentseparations, we find that all 9 are likely to be foundin the range 0.′′33 to 1.′′5. Of the 50 binaries in therange 150-675 AU we thus should count only 41 asphysical binaries, leading to the abovementionedbinary percentage of 5.3%. We thus find that thebinary fraction in associations is higher by a factorof 2.2 compared to the ONC, in qualitative agree-ment with the study of Petr et al. (1998), whofound a difference of almost a factor of 3 based onsmall-number statistics.

4.2. Separation Distribution Function

In Figure 6 we show the separation distributionof binaries towards the ONC on an angular scaleand in bins 0.′′1 wide. The first bin from 0.′′1 to0.′′2 is incomplete. It is striking that there is clearevidence for structure in the separation distribu-tion function, with a sudden decrease in the num-ber of binaries as the separations increase beyond0.′′5, corresponding to 225 AU. At larger separa-tions, the distribution is quite flat. We return tothe physical interpretation of this “wall” in subse-quent sections. An early precursor to this resultmay have been seen by Simon (1997), who stud-ied the two-point correlation function in the ONCbased on the results of Prosser et al. (1994) andnoted a transition from binary companions to thelarge-scale cluster at projected separations around400 AU.

We have made the same plot on a logarithmicscale in Figure 7. The data are essentially com-plete over the whole separation range displayed.We have corrected the distribution for the 9 line-of-sight pairs by calculating the probability of find-ing another star within a circle with radius corre-sponding to the separations of each binary, andthen adding these probabilities up for each bin.While such a procedure is meaningless for an in-dividual object, it can be used to distribute the 9false binaries across the 5 bins. Binaries with thelargest separations are most likely to be line-of-

sight pairs, and the last bin has a large correctionof 6 stars, while three of the four remaining binseach is corrected for one star. Error bars are givenfor each corrected column.

Figure 7 additionally shows the distribution offield stars in the same interval from Duquennoy &Mayor (1991). The dot-dash curve represents theGaussian that Duquennoy & Mayor fitted to theirentire data set, while the two dashed crosses rep-resent their actual data points within our range.It is evident that the Gaussian is a poor fit tothese data points in this specific separation inter-val. The Duquennoy & Mayor data are given as afunction of period, and we have converted these toseparations by assuming a mean total binary massof 1.2 M⊙, appropriate for their sample of F7 to G9binaries. We have also assumed, as is commonlydone, that the projected separation represents thesemimajor axis, although statistically there is asmall difference (Kuiper 1935, Couteau 1960).

As noted above, we find a binary frequency(after correcting for the 9 line-of-sight pairsand treating the three triples as six binaries) of8.8%±1.1% in the interval from 67.5 AU to 675AU. If we calculate the binary frequency fromthe Duquennoy & Mayor data in the same range,we find a binary frequency of 13.7% using a sim-ple trapezoidal approximation to their log-normalcurve, and 12.4% using linear interpolation of theirdata points. It follows that, in the specific rangefrom 67.5 AU to 675 AU, there is approximately1.5 times more binaries in the field than in theONC. We discuss this result further below.

4.3. Wide vs Close Binaries as Function of

Distance from the Trapezium

We have explored whether there is a differencein the number of wide binaries relative to closebinaries as we move from the dense inner clusterregions to the outer reaches of the cluster. Sucha difference has been searched for but not foundin earlier studies (Kohler et al. 2006). In view ofthe dramatic change in the separation distributionfunction at 0.′′5, we have chosen this separation asa dividing point, such that we consider binariesbetween 0.′′15 and 0.′′5 as ‘close’ and binaries be-tween 0.′′5 and 1.′′5 as ‘wide’. Figure 8 shows thecumulative distributions of these close and widebinaries as function of distance from θ1 Ori C inthe Trapezium. It is evident from the figure that

5

the two sets of binaries do not have the same ra-dial distribution with increasing distance from thecluster core. To explore this further, we show inFigure 9 the ratio of wide to close binaries as acontinuous cumulative distribution, with the firstpoint calculated 30′′ outside the exclusion zone,that is, at a distance of 90′′ from θ1 Ori C. For eachdistance, the curve gives the ratio for all binariesfrom the edge of the exclusion zone to that partic-ular distance. In other words, as the curve movesaway from the Trapezium it accounts for more andmore binaries, until the last points on the curverepresent the mean ratio of wide-to-close binariesfor the entire ONC. The dashed lines indicate the1σ errors on the numbers. We have furthermorecalculated the same ratio for the Duquennoy &Mayor (1991) binaries, and the two dotted linesindicate the values calculated from their Gaussianfit (lower line) and their actual data points (upperline). These lines thus represent the ratio for fieldstars.

The curve does not take into account that 9 ofthe binaries are likely to be line-of-sight associa-tions. Given that the probability of being a line-of-sight pair increases with separation and withproximity to the center of the ONC, it follows thatthe curve would dip even deeper down in the innerregion if we could remove the 9 non-physical pairs.

It is evident that there is a very pronounced andalmost monotonic change in the ratio of wide toclose binaries as one moves away from the core ofthe ONC until a distance of about 460′′, at whichpoint the ratio becomes flat. It is also clear thatthe mean ratio of wide to close binaries for thewhole ONC is lower than the Duquennoy & Mayorvalues. We discuss the implications of these find-ings in Section 5.

4.4. Higher-order Multiples

In addition to the 72 binaries in the separa-tion range from 0.′′15 to 1.′′5, we have found 3triple systems with separations in the same range.Tokovinin (2001) suggests that the ratio of triplesto binaries is 0.11±0.04 in general, whereas af-ter subtracting the 9 line-of-sight pairs we find0.048±0.01 in the ONC. Within the considerableerrors the numbers are close to being consistent,but it should also be recalled that our survey isrestricted to the limited range between the 0.′′15completeness limit and the 1.′′5 confusion limit. It

is entirely possible that some close pairs in triplesystems have already evolved to become spectro-scopic binaries and have thus become unobserv-able with our detection method. Nor can it beexcluded that some wide pairs exist in triple sys-tems with separations larger than 1.′′5, althoughsuch wide systems cannot be common (Scally etal. 1999). Much better statistics is required tosettle the question whether the ONC might be de-ficient in triple systems.

4.5. Flux Ratios and the Nature of Com-

panions

We have determined the flux-ratios of those ofour binaries that are not saturated in our images.Figure 10 shows the resulting histogram as a func-tion of ∆mag. As is well established from other bi-nary studies, the majority of binaries in the ONCalso have unequal components, although the binwith equal components (∆m < 0.5 mag.) is themost populated. Only very few of the companionshave a magnitude difference to their primary ofmore than 2.5 magnitudes.

In the absence of spectroscopic information, wehave made a crude attempt to investigate the na-ture of the companions based on their measuredflux ratios. In order to do that we have madesome assumptions. First, we assume that the ob-served flux ratios reflect the photospheric fluxes,in other words that Hα emission line fluxes arenot seriously affecting the ratios. This may notalways be a good assumption, since mid-to-lateM dwarfs often are very active, and their photo-spheric fluxes are so low that Hα line emissioncould be a significant contribution. If we mis-take Hα line emission for photospheric flux fromthe primary, our estimate of a spectral type fora companion will be earlier than it is in reality.Second, most published photometry of late-typedwarfs is broad-band, whereas we have observedin the narrow-band Hα filter, so we assume thatour observed magnitude differences can be com-pared to R-band photometry. Since we are deal-ing with the difference between two stars, this isprobably not a bad assumption, at least when theflux ratio is not large. Third, we assume that theobserved flux ratios are not affected by differencesin extinction between the components. Given theyoung age and occasional association with molec-ular clouds, this may not in all cases hold true.

6

With these caveats spelled out, we have usedthe MI vs spectral type relation and the R-I col-ors for M dwarfs (e.g., Dahn et al. 2002) to derivethe difference in R-band magnitudes as a functionof spectral type for M-dwarfs. The relation turnsout to be essentially linear, with a mean drop of0.56 magnitudes per spectral subtype throughoutthe M spectral range. We then used the spectralclassifications for the primaries provided by SIM-BAD, of mixed provenance, with our flux ratios toestimate a spectral type for the secondaries. Fora cluster with an age between 0.5 and 3 Myr, thesubstellar limit is around spectral type M6.25 fol-lowing the models of Baraffe et al. (1998) andChabrier et al. (2000) and using the temperaturescale of Luhman et al. (2003), see also Luhmanet al. (2006). To our surprise, quite a numberof the secondaries appear to be brown dwarfs, inat least one case forming a wide BD-BD binary.We comment on selected cases in the Appendix.Given the assumptions involved and the simplisticnature of these spectral type estimates, it is obvi-ous that spectroscopy is required to establish thetrue nature of the secondaries.

5. DISCUSSION

5.1. Structure and Evolution of the Sepa-

ration Distribution Function

The separation distribution function of field bi-naries has been approximated by various func-tions. Opik (1924) suggested that binaries withseparations larger than ∼60 AU follow a f(a)∼ 1/a distribution, whereas Kuiper (1942) pro-posed that the overall distribution could be repre-sented by a Gaussian. The latter distribution wasadopted as a good fit to their data in the influentialstudy by Duquennoy & Mayor (1991), although itis evident from Figure 7 that in the specific sep-aration range discussed in the present paper, theGaussian is a poor approximation to the actualdata. Integrating the Opik relation over logarith-mic bins results in a straight horizontal line to de-scribe the logarithmic separation distribution, andthis is clearly a better fit to the two data pointsof Duquennoy & Mayor seen in Figure 7. TheOpik relation has also been supported by numer-ous studies summarized by Poveda & Allen (2004).

For the separation distribution of ONC bina-ries (corrected for line-of-sight associations) both

a Gaussian and a horizontal line provide accept-able fits to the data when considering the uncer-tainties of the data points. It is thus not possibleto classify the structure of the ONC separationdistribution function with the available data.

The separation distribution function in theONC most probably has not yet found its finalshape. Two basic mechanisms operate that canaffect the orbit of a young binary: rapid dynamicaldecay in small-N clusters (e.g., Sterzik & Durisen1998, Reipurth & Clarke 2001, Bate et al. 2002)and the passage of a binary through a dense clus-ter (e.g., Kroupa 1995a,b, Kroupa et al. 1999).The former occurs primarily during the Class 0phase (Reipurth 2000), and is no longer relevantfor stars in the ONC. The latter, however, maybe essential for understanding the binary popula-tion of the ONC. A wide binary falling throughthe potential well of a cluster will gain kinetic en-ergy through encounters, and binaries with weakbinding energies are eventually disrupted (Heggie1975). A prediction of this scenario is that binariesat distances from the cluster center larger than thecorresponding crossing time should not be show-ing any dynamical alterations due to encounterswith other cluster members (Kroupa et al. 1999,2001). The crossing time of a star through a clus-ter is tcross = 2R/σ, where R is the cluster radiusand σ is the mean one-dimensional velocity dis-persion in the cluster. Assuming that this velocitydispersion in the ONC is of the order of 2 km s−1

(e.g., Jones & Walker 1988), we have in Figure 9indicated the crossing time for different distancesto the cluster center. The figure suggests that fordistances larger than roughly 460 arcsec there is nolonger a measurable change in the ratio of wide-to-close binaries. Given that the wide-to-close binaryratio changes by a factor of 4-5 from the inner tothe outer regions, this suggests that many, andperhaps most, of the wide binaries are disruptedafter only a few passages of the cluster center.The variation of the ratio of wide-to-close binariesfrom the inner to the outer regions of the ONC,seen in Figure 9, offers the first compelling obser-vational evidence that dynamical interactions inthe dense central region of the ONC have takenplace.

In principle, the diagram in Figure 9 allows usto determine the age of the ONC. An angular dis-tance of ∼460 arcsec corresponds to a crossing

7

time of about 1 million years. However, an agedetermined in this manner is directly dependenton the velocity dispersion assumed. Therefore, allwe can say about the age of 1 million years weestimate for the ONC is that it is consistent withother ONC age estimates (e.g., Hillenbrand 1997).

The significantly lower number of binaries thatwe find in the ONC compared to associations thusappears to be due, at least in part, dissolutionof wide binaries. However, under certain circum-stances an encounter could lead to hardening ofthe binary, making it closer than our resolutionlimit, so it is not lost to the overall binary budget.

Even in its outermost regions, the ONC showsa smaller ratio of wide-to-close binaries than seenin the field by Duquennoy & Mayor (1991). Con-sidering that the majority of field stars are likelyto have been formed in a cluster, it follows thatmany of the wide binaries in the field must haveformed in the gentler environment of a loose Tassociation.

Durisen & Sterzik (1994) found, on theoreticalgrounds, that binaries are more likely to form inclouds with lower temperatures. Reipurth & Zin-necker (1993) found observational evidence thatclouds with more stars have relatively fewer bina-ries in the separation range under study (mostlywide visual binaries). Both these results couldindicate that loose T associations produce or re-tain more wide binaries than do clusters. How-ever, Brandeker et al. (2006) noted that the youngsparse η Chamaeleontis cluster has a deficit of widebinaries. Unless this small cluster is the remnantof a much denser cluster, then this result wouldseem to be in contradiction to the notion thatwide binaries are preferentially formed/preservedin loose associations.

In any case, there is no question that the fieldbinary population is a mix of binaries formed inclusters and in associations. We can attempt tocalculate the fraction of stars that originate inclusters and in associations. In Sect. 4.2 weshowed that, in the limited separation range 67.5- 675 AU, the binary fraction of the field stars ofDuquennoy & Mayor (1991) is about a factor 1.5higher than in the ONC. We also found that inthe range 150 - 675 AU associations have a factor2.2 more binaries than the ONC. In other words,in these limited ranges associations have about 1.5times as many binaries as the field population, and

the ONC has only about 2/3 as many binaries asthe field population. It follows that the field starbinary population in principle can be produced ifonly 1/3 of binaries come from the binary-rich as-sociations and 2/3 originate in binary-poor ONC-type clusters, in excellent agreement with the re-sults of Patience et al. (2002) and in approxi-mate agreement with findings that 70-90% of allstars may be formed in clusters (e.g., Lada & Lada2003). Of course, this assumes that the binary ra-tios between field stars, associations, and the ONCderived here within narrow separation ranges canbe extrapolated to all separations, an assumptionthat may not hold at all.

5.2. Very Low Mass Stellar and Substellar

Companions

As noted in Sect. 4.5, spectroscopy is requiredto determine the true nature of the companionswe have found. However, unless significant ex-tinction differences are common among our bina-ries, then the difference in component brightness,combined with spectral type estimates for the pri-maries when available, are indicative of the spec-tral types of the secondaries. While this is highlyuncertain in individual cases, overall it is likely tobe indicative of the properties of the companionpopulation. Table 1 lists the spectral types for46 primaries, and of these 38, or 83%, are M-typestars. We find that more than half (25) of these 46binary stars have secondaries with spectral typesof M5 or later, and about one third (17) could besubstellar. These numbers are upper limits, sincewe cannot correct for the 9 false binaries that weexpect to be present in our sample of 78 multi-ple systems, but they are probably more likely tobe found among pairs with faint, widely separatedcompanions. However, even if we subtracted all 9objects from the 17 that could be substellar, thiswould still indicate that 8, or about 17%, of the46 binaries with spectral type estimates are likelyto have substellar companions. Selected cases arediscussed in the Appendix.

The closest binaries we were able to resolve withour imaging technique have projected separationsof 50 AU. All of the binaries with likely substel-lar companions are thus quite wide, which is ofinterest for formation theories (see, e.g., Lucas etal. 2005, and reviews by Burgasser et al. 2007,Luhman et al. 2007, Whitworth et al. 2007).

8

It is likely that there are several brown dwarf –brown dwarf binaries in our binary sample, but theonly one that we are certain of is COUP 1061. In-frared spectroscopy of this source indicates a spec-tral type of M9-L0 (Meeus & McCaughrean 2005).We have resolved this object into two componentswith almost equal brightness and a projected sep-aration of 100 AU. Brown dwarf binaries with suchlarge separations are rare (e.g., Lucas et al. 2005,Allen et al. 2007), although a few very wide pairsare known (e.g., Artigau et al. 2007, Barrado yNavascues et al. 2007).

6. CONCLUSIONS

We have analyzed a large set of Hα images ofthe Orion Nebula Cluster acquired with the Hub-ble Space Telescope and the Advanced Camera forSurveys with the goal of detecting binaries amonga sample of 781 ONC members. The following re-sults were obtained:

1. A total of 75 binaries and 3 triple systemswere detected, of which 55 are new discoveries.

2. Within the limited angular range of 0.′′15to 1.′′5, corresponding to projected separations of67.5 AU to 675 AU, we have found a binary frac-tion of 8.8%±1.1% after correcting for a statisti-cally determined contamination of 9 line-of-sightbinaries.

3. The field binary fraction for solar type starsin the same separation range is 1.5 times larger,and for T Tauri associations it is 2.2 times largerthan in the ONC, confirming earlier results thatthe ONC is deficient in binaries, now with statis-tically significant data.

4. The separation distribution function foryoung binaries in the ONC shows a dramatic de-crease in binaries for angular separations largerthan 0.5′′, corresponding to projected separationsof 225 AU.

5. The ratio of cumulative distributions of wide(0.′′5 to 1.′′5) to close (0.′′15 to 0.′′5) binaries showan increase out to a distance of about 460′′ fromthe center of the ONC, after which it levels out.We interpret this as clear observational evidencefor dynamical evolution of the binary populationas a result of passages through the potential wellof the ONC. These results are consistent with anage of the ONC of about 1 million yr.

6. It appears that much of the deficiency of

binaries in the ONC compared to the field starpopulation can be understood, at least in part, interms of the destruction of wide binaries combinedwith a secondary effect from orbital evolution ofbinaries towards closer separations that are unob-servable in direct imaging surveys.

7. Limited spectral information about the pri-mary stars indicate that they are low mass T Tauristars (except for one Herbig Ae/Be star). Assum-ing that most binaries are not affected by differ-ential extinction, we find that possibly as many as50% of the binaries have companions with spec-tral types later than M5, and from 1/6 to 1/3 ofthe binaries may have substellar companions, allof which with separations of at least 50 AU. Thislarge number of wide substellar companions is ofinterest for theories of brown dwarf formation.

We thank Rainer Kohler, Pavel Kroupa, LuizPaulo Vaz, and Adam Burgasser for valuable com-ments, and an anonymous referee for a very help-ful referee report. We are grateful to BurtonJones for providing us an electronic version ofthe Jones & Walker (1988) proper motion table.Marcelo Guimaraes acknowledges financial sup-port by CAPES, CNPq and FAPEMIG. Based onobservations taken under program GO-9825 withthe NASA/ESA Hubble Space Telescope obtainedat the Space Telescope Science Institute, whichis operated by the Association of Universities forResearch in Astronomy, Inc., under NASA con-tract NAS5-26555. This material is based uponwork supported by the National Aeronautics andSpace Administration through the NASA Astro-biology Institute under Cooperative AgreementNo. NNA04CC08A issued through the Office ofSpace Science. This research has made use of theSIMBAD database, operated at CDS, Strasbourg,France, and NASA’s Astrophysics Data SystemBibliographic Services.

9

A. Appendix. Individual Binaries of Interest.

In the following, comments are provided on binaries of particular interest, especially regarding possiblesubstellar companions.

V1438 Ori (JW 39). The spectral type of M3 is due to Hillenbrand (1997). If the 2.6 mag brightnessdifference is not affected by extinction differences, then the companion is an M8 brown dwarf at a projectedseparation of 170 AU. Stassun et al. (1999) detected lithium in the primary.

JW 71. The spectral type of M4 is due to Hillenbrand (1997). If the 1.4 mag brightness difference is notaffected by extinction differences, then the companion is an M7 brown dwarf with a projected separation of90 AU.

V1118 Ori (JW 73). This is a member of the class of EXor’s (Herbig 1989). The star, which is alsoknown as Chanal’s Object, has had 5 outbursts since its discovery in 1984, and typically varies within therange 14.5 <V<18, with a rise time of less than a year and a declining phase about twice as long. Optical andinfrared spectroscopy of V1118 Ori is reported by Parsamian et al. (2002) and Lorenzetti et al. (2006), andX-ray observations are analyzed by Audard et al. (2005). The discovery that V1118 Ori has a companiononly 0.4 magnitude fainter in the Hα filter raises interesting questions about which of the components thatmay drive the variability (Herbig 2007).

JW 121. The spectral type of M3 is due to Hillenbrand (1997). If the 3.0 mag brightness difference isnot affected by extinction differences, then the companion is an M8 brown dwarf with a projected separationof 153 AU.

JW 147. The spectral type of M5 is due to Hillenbrand (1997). If the 1.4 mag brightness difference isnot affected by extinction differences, then the companion is an M8 brown dwarf with a projected separationof 135 AU.

JW 152. The spectral type of M3 is due to Hillenbrand (1997). If the 2.5 mag brightness difference isnot affected by extinction differences, then the companion is an M7 brown dwarf with a projected separationof 81 AU.

JW 235. JW 235 is associated with the HH 504 object, discovered by Bally & Reipurth (2001), see alsoO’Dell (2001). The binarity of the star was discovered by Kohler et al. (2006). Its proper motion in theJones & Walker catalog implies that it has 0% probability of being a member of the ONC. However, giventhe similar brightness of the two components, it is likely that brightness variations of the stars shifted thephotocenter used for astrometry. Evidence for ONC membership is strong and is based on the presence ofHα emission, detection in X-rays, optical variability, and association with an HH object.

V1274 Ori (JW 248). The spectral type of the primary is somewhat in dispute. Hillenbrand (1997)suggests M0.5-M2 (with the Ca II lines in emission), Edwards et al. (1993) suggest M3, and Meeus &McCaughrean (2005) suggest M0-M4. The binarity of the system was found by the COUP survey. Ourprimary is the secondary in the COUP catalog, indicating that our secondary is highly X-ray active. Sicilia-Aguilar et al. (2005) detected lithium in the (optical) primary, and found an inverse P Cygni profile at Hα.If there is not an extinction difference between the primary and secondary, then the large flux difference(>3.7 mag) suggests that the companion could be an L0 brown dwarf. However, the system is only 3 arcminfrom θ1 Ori C, so there is a 1% chance of a line-of-sight association.

JW 296. JW 296 is the primary of a hierarchical triple system. The secondary and tertiary are veryfaint stars, surrounded by a common proplyd-like envelope known as 066-652 (O’Dell & Wong 1996). Theprimary is of spectral type M4.5 according to Hillenbrand (1997). If the brightness difference is taken atface value, the secondary and tertiary should be L0 brown dwarfs, but given the obvious association withnebulosity it is likely that their faintness is merely due to extinction.

JW 355. The primary of this system is located at the edge of a proplyd known as 109-246 (O’Dell &Wong 1996). The proper motion of the star suggests it is not an ONC member, but it is found to be amember by the COUP project. Hillenbrand (1997) suggests a mid-K spectral type. The system is only 90′′

from θ1 Ori C, and the star density is so high that the possibility of a line-of-sight association is several

10

percent.

JW 370. There is a silhouette disk close (∼3.′′4) to this binary. Given the local density of stars aroundthis system, the probability that such a silhouette disk is just a chance alignment is ∼7%. The chance of aline-of-sight alignment for the binary itself is 1.5%. Hillenbrand (1997) suggests a spectral type of K0-K3.

V1492 Ori (JW 383). The primary is of spectral type M3 according to Hillenbrand (1997). If the 2.1mag brightness difference is not affected by extinction differences, then the companion is an M7 brown dwarfwith a projected separation of 86 AU. Stassun et al. (1999) derive a rotation period of 7.11 days for theprimary, and notes the presence of lithium in the spectrum.

Parenago 1806 (JW 391). The primary is of spectral type M1 according to Edwards et al. (1993) andHillenbrand (1997), who also notes the presence of Ca II emission. The primary is saturated in our images,but with a magnitude difference of at least 3.3 mag, the companion would be an M7 brown dwarf if there isno difference in extinction. However, the pair is located in a region of high stellar density, so the propabilityof a chance alignment is more than 5%.

JW 509. This binary was first detected by Prosser et al. (1994) and subsequently by Padgett et al.(1997) and Lucas et al. (2005). Although it is a well known binary it does not have a spectral type in theliterature. Since this system is located in a crowded region, the probability of a chance alignment is ∼ 2.7%,but the reality of the binary is supported by an interaction zone between the stars, visible in our image (Fig.3b).

Parenago 1914 (JW 551). The primary of this hierarchical triple system has a spectral type M1 accordingto Hillenbrand (1997). The tertiary is not detectable either in the optical or in X-rays, and since its separationfrom the primary is as large as 1.′′27 it could be a background object. The system seems to be part of a smallsubcluster, which gives it a high probability of a chance alignment (∼ 2.2%). Given the primary spectraltype, the difference in magnitudes to the tertiary (3.6) and assuming that the tertiary is not a backgroundobject, the tertiary could have a spectral type M7 or M8 and thus would be a brown dwarf candidate.

COUP 967. This star was catalogued as a binary by Prosser et al. (1994), Padgett et al. (1997) andLucas et al. (2005). Our image shows that the primary is associated with a proplyd (184-427 in O’Dell& Wong 1996) and the secondary is a faint object (∆mHα = 2.4). The spectral type for the primary isM2.5 according to Hillenbrand (1997). The system is very close to θ1 Ori C (∼ 70′′) and the probability ofa chance alignment is 3.5%. If the components have the same extinction, then the secondary would be ofspectral type M6.5, and thus at the hydrogen burning limit.

JW 592. The spectral type M2.5 is due to Hillenbrand (1997) and M4 according to Edwards et al. (1993).The brightness difference (4 magnitudes) would imply an M9.5 spectral type or later for the secondary,making it a brown dwarf with a projected separation of ∼ 275 AU.

COUP 1061. This very close system (∼ 100 AU) has a combined spectral type M9-L0 according toMeeus & McCaughrean (2005), who did not resolve it, and since the components have virtually the samebrightness, this is therefore a bona fide brown dwarf - brown dwarf binary with a projected separation of100 AU.

V1524 Ori (JW 681). This binary was first catalogued by Prosser et al. (1994) and has a spectral typeK7 according to Prosser & Stauffer (unpublished). The secondary is associated with the proplyd 213-346(O’Dell & Wong 1996). In the Hillenbrand (1997) catalog this system has two entries, both with number681 but with different coordinates. SIMBAD identifies three different objects: H97b-681, H97b-681a, andH97b-681b. The first is JW 681, also known as V1524 Ori, and is the primary in this double. The second isthe unrelated object MLLA 312, and the third is the X-ray source COUP 1149, which forms the secondarycomponent, associated with the proplyd. Because this system is located near to θ1 Ori C (∼ 77′′) theprobability of a chance alignment is 3.5%.

V1528 Ori (JW 727). The spectral type M2 is due to Hillenbrand (1997) and the difference in brightnesssuggests a spectral type of L0 for the companion if the components have the same extinction. If so, thesecondary is a brown dwarf with projected separation of 329 AU. The primary shows Hα emission according

11

to our unpublished survey.

JW 748. This binary was discovered by Prosser et al. (1994), and the spectral type of the primaryis M3 according to Hillenbrand (1997). Given the brightness diffference and if the components have thesame extinction, the secondary could be a brown dwarf with spectral type M7 and a projected separation of158 AU.

JW 767. First discovered by Kohler et al. (2006) this system has a spectral type M2.5 due to Hillenbrand(1997). The secondary has a silhoutte disk and is associated with the HH 668 object (Bally et al. 2006).The extinction caused by the silhoutte disk probably is responsible for the large difference in brightness bymore than 4 magnitudes between the primary and secondary.

Parenago 2075 (JW 867). This object was observed by Kohler et al. (2006) but despite its brightnessand separation (0.′′29) they did not resolve it, suggesting possible major variability of the secondary. Thesystem presents evidence of Hα emission, X-rays and optical variability. A spectral type of M1 was suggestedby Blanco (1963), and more recently Duncan (1993) assigned a spectral type of K8Ve. Sicilia-Aguilar et al.(2005) reported lithium in the spectrum of the primary.

JW 906. The primary is of spectral type M3 according to Hillenbrand (1997), and if the difference inbrightness (2.7 mag) is not affected by extinction then the secondary is a brown dwarf with spectral type M8and a large projected separation of 625 AU. Sicilia-Aguilar et al. (2005) reported lithium in the spectrumof the primary.

12

REFERENCES

Allen, P.R., Koerner, D.W., McElwain, M.W.,Cruz, K.L., & Reid, I.N. 2007, AJ, 133, 971

Artigau, E., Lafreniere, D., Doyon, R., Albert, L.,Nadeau, D., & Robert, J. 2007, ApJ, 659, L49

Audard, M., Gudel, M., Skinner, S.L., Briggs,K.R., Walter, F.M. et al. 2005, ApJ, 635, L81

Bally, J. & Reipurth, B. 2001, ApJ, 546, 299

Bally, J., Licht, D., Smith, N., & Walawender, J.2006, AJ, 131, 473

Baraffe, I., Chabrier, G. Allard, F., & Hauschildt,P.H. 1998, A&A, 337, 403

Barrado y Navascues, D., Bayo, A., Morales-Calderon, M., Huelamo, N., Stauffer, J.R., &Bouy, H. 2007, a, in press

Bate, M.R., Bonnell, I.A., & Bromm, V. 2002,MNRAS, 336, 705

Blanco, V.M. 1963, ApJ, 137, 513

Brandeker, A., Jayawardhana, R., Khavari, P.,Haisch, K.E., & Mardones, D. 2006, ApJ, 652,1572

Burgasser, A.J., Reid, I.N., Siegler, N., Close, L.,Allen, P., Lowrance, P., & Gizis, J. 2007, inProtostars and Planets V, eds. B. Reipurth, D.Jewitt, & K. Keil, Univ. of Arizona Press, 427

Chabrier, G., Baraffe, I., Allard, F., & Hauschildt,P.H. 2000, ApJ, 542, 464

Choi, P.I. & Herbst, W. 1996, AJ, 111, 283

Correia, S., Zinnecker, H., Ratzka, Th., & Sterzik,M.F. 2006, A&A, 459, 909

Couteau, P. 1960, Journ. des Observateurs, 43, 41

Dahn, C.C., Harris, H.C., Vrba, F.J., Guetter,H.H., Canzian, B., Henden, A.A. et al. 2002,AJ, 124, 1170

Duchene, G. 1999, A&A, 341, 547

Duchene, G., Bouvier, J., Bontemps, S., Andre,P., & Motte, F. 2004, A&A, 427, 651

Duncan, D.K. 1993, ApJ, 406, 172

Duquennoy, A. & Mayor, M. 1991, A&A, 248, 485

Durisen, R.H. & Sterzik, M.F. 1994, A&A, 286, 84

Edwards, S., Strom, S.E., Hartigan, P., Strom,K.M., Hillenbrand, L.A. et al. 1993, AJ, 106,372

Getman, K.V., Feigelson, E.D., Grosso, N., Mc-Caughrean, M.J., Micela, G. et al. 2005, ApJS,160, 353

Heggie, D.C. 1975, MNRAS, 173, 729

Herbig, G.H. 1962, Adv. Astr. Astrophys., 1, 47

Herbig, G.H. 1982, in Symp. on the Orion Nebulato Honor Henry Draper, ed. A.E. Glassgold &P.J. Huggins, New York Acad. Sci., 64

Herbig, G.H. 1989, in ESO Workshop on LowMass Star Formation and Pre-Main SequenceObjects, ed. Bo Reipurth, p.233

Herbig, G.H. 2007, in prep.

Herbig, G.H. & Terndrup, D.M. 1986, ApJ, 307,609

Hillenbrand, L.A. 1997, AJ, 113, 1733

Jones, B.F. & Walker, M.F. 1988, AJ, 95, 1755

Joy, A.H. & van Biesbroeck, G. 1944, PASP, 56,123

Kroupa, P. 1995a, MNRAS, 277, 1491

Kroupa, P. 1995b, MNRAS, 277, 1507

Kroupa, P. 2000, New Astronomy, 4, 615

Kroupa, P., Petr, M.G., & McCaughrean, M.J.1999, New Astronomy, 4, 495

Kroupa, P., Aarseth, S., & Hurley, J. 2001, MN-RAS, 321, 699

Kuiper, G.P. 1935, PASP, 47, 121

Kuiper, G.P. 1942, ApJ, 95, 201

Kohler, R., Petr-Gotzens, M. G., McCaughrean,M. J., Bouvier, J., Duchene, G. et al. 2006,A&A, 458, 461

Lada, C.J. & Lada, E.A. 2003, Ann. Rev. Astron.Astrophys. 41, 57

13

Lorenzetti, D., Giannini, T., Calzoletti, L., Puc-cetti, S., Antoniucci, S. et al. 2006, A&A, 453,579

Lucas, P.W., Roche, P.F., & Tamura, M. 2005,MNRAS, 361, 211

Luhman, K.L., Stauffer, J.R., Muench, A.A.,Rieke, G.H., Lada, E.A., Bouvier, J., & Lada,C.J. 2003, ApJ, 593, 1093

Luhman, K.L., Whitney, B.A., Meade, M.R.,Babler, B.L., Indebetouw, R., Bracker, S., &Churchwell, E.B. 2006, ApJ, 647, 1180

Luhman, K.L., Joergens, V., Lada, C., Muzerolle,J., Pascucci, I., & White, R. 2007, in Protostarsand Planets V, eds. B. Reipurth, D. Jewitt, &K. Keil, Univ. of Arizona Press, 443

Meeus, G. & McCaughrean, M.J. 2005, AN, 326,977

O’Dell, C.R. 2001, AJ, 122, 2662

O’Dell, C.R. & Wong, K. 1996, AJ, 111, 846

Padgett, D.L., Strom, S.E., & Ghez, A. 1997, ApJ,477, 705

Parsamian, E.S., Mujica, R., & Corral, L. 2002,Astrophysics, 45, 393

Patience, J., Ghez, A.M., Reid, I.N., & Matthews,K. 2002, AJ, 123, 1570

Petr, M.G., Coude du Foresto, V., Beckwith, S.,Richichi, A., & McCaughrean, M.J. 1998, ApJ,500, 825

Poveda, A. & Allen, C. 2004, Rev. Mex. Astron.Astrofis. Ser. Conf. 21, 49

Preibisch, T., Balega, Y., Hofmann, K.-H.,Weigelt, G., & Zinnecker, H. 1999, New As-tron., 4, 531

Prosser, C.F., Stauffer, J.R., Hartmann, L.,Soderblom, D.R., Jones, B.F. et al. 1994, ApJ,421, 517

Ratzka, T., Kohler, R., & Leinert, Ch. 2005, A&A,437, 611

Reipurth, B. 2000, AJ, 120, 3177

Reipurth, B. & Clarke, C.J. 2001, AJ, 122, 432

Reipurth, B. & Zinnecker, H. 1993, A&A, 278, 81

Scally, A., Clarke, C.J., & McCaughrean, M.J.1999, MNRAS, 306, 253

Sicilia-Aguilar, A., Hartmann, L.W., SzentgyorgyiA.H., Fabricant, D.G., Furesz, G. et al. 2005,AJ, 129, 363

Simon, M. 1997, ApJ, 482, L81

Simon, M., Ghez, A. M., Leinert, Ch., Cassar, L.,Chen, W. P. et al. 1995, ApJ, 443, 625

Simon, M., Close, L.M., Beck, T.L. 1999, AJ, 117,1375

Stassun, K.G., Mathieu, R.D., Mazeh, Tsevi, &Vrba, F.J. 1999, AJ, 117, 2941

Sterzik, M.F. & Durisen, R.H. 1998, A&A, 339, 95

Tokovinin, A. 2001, in IAU Symp. No.200 TheFormation of Binary Stars, eds. H. Zinnecker& R.D. Mathieu, p.84

Opik, E. 1924, Tartu Obs. Publ., 25, No. 6

Whitworth, A., Bate, M.R., Nordlund, A.,Reipurth, B., & Zinnecker, H. 2007, in Proto-stars and Planets V, eds. B. Reipurth, D. Je-witt, & K. Keil, Univ. of Arizona Press, 459

This 2-column preprint was prepared with the AAS LATEXmacros v5.2.

14

Table 1

Pre-main sequence binaries with angular separation between 0.′′15 and 1.′′5, outside the 60′′

exclusion zone around θ1 Ori C.

JW % Par GCVS COUP H97 RA2000 DEC2000 PA SEP [′′] mI ∆mHα1 PAOriC OriC SpT2 Memb.3 Bin

39 99 · · · V1438 · · · 39 5:34:38.1 -5:27:41 21 0.38 14.29 2.6 246 628 M3 PV52 98 · · · · · · · · · 52 5:34:40.8 -5:28:09 231 0.20 14.58 0.1 242 604 M5 P63 99 1569 V1441 17 63 5:34:43.0 -5:20:07 0 0.27 12.75 1.5 291 537 K6 PXV71 99 · · · · · · 21 71 5:34:44.5 -5:24:38 105 0.20 15.61 1.4 261 483 M4 PX73 99 · · · V1118 · · · 73 5:34:44.7 -5:33:42 329 0.18 14.04 0.4 217 779 M1 PHV81 97 1600 · · · 28 81 5:34:46.4 -5:24:32 272 1.34 12.91 1.9 261 454 M0 PXV121 99 · · · · · · · · · 121 5:34:51.2 -5:16:55 200 0.34 14.90 3.0 316 541 M3 PV124 99 · · · · · · 64 124 5:34:51.8 -5:21:39 200 0.48 13.94 0.0 286 382 M3.5 PXV127 99 · · · · · · 66 127 5:34:52.1 -5:24:43 254 0.48 14.28 0.6 258 373 M3.5 PXHV128 97 · · · V1458 67 128 5:34:52.2 -5:22:32 215 0.39 12.64 >0.6 278 366 M2.5 PXHV135 99 · · · V1460 72 135 5:34:52.7 -5:29:46 11 0.40 14.45 0.1 223 522 M3 PXV147 99 · · · · · · · · · 147 5:34:54.4 -5:17:21 26 0.30 14.80 1.4 318 489 M5 P151 99 · · · · · · 95 151 5:34:54.8 -5:25:13 332 0.15 15.44 1.6 251 341 M3 PX152 99 · · · · · · 96 152 5:34:55.1 -5:25:30 51 0.18 14.12 2.5 248 343 M3 PX176 99 1673 KQ 123 176 5:34:57.8 -5:23:53 336 1.28 11.58 >1.1 264 280 K8 PXV190 99 · · · · · · 134 190 5:34:59.3 -5:23:33 355 0.56 15.33 0.3 268 256 M6 PX201 99 · · · · · · 150 201 5:35:01.0 -5:24:10 246 1.09 13.41 0.8 258 235 M2.5 PXH222 98 · · · V1320 174 222 5:35:02.2 -5:29:10 89 1.07 14.14 1.4 212 407 M2 PXV223 99 · · · · · · 177 223a 5:35:02.4 -5:20:47 130 0.34 13.64 1.3 307 261 · · · PX224 99 · · · · · · 180 224 5:35:02.7 -5:19:45 164 0.27 15.38 2.4 317 299 M1 PXV235 0 · · · · · · 197 235 5:35:03.6 -5:29:27 346 0.54 13.69 0.4 208 411 · · · PXHV248 99 · · · V1274 214 248 5:35:04.4 -5:23:14 320 0.90 12.73 >3.7 273 180 M3 PXV- - · · · · · · 260 3064 5:35:06.2 -5:22:13 284 0.40 16.20 0.3 295 169 M4 X

296 99 · · · · · · 275 296 5:35:06.6 -5:26:51 196 0.89 14.30 3.2 215 255 M4.5 PX296 99 · · · · · · 275 296 5:35:06.6 -5:26:52 158 0.27 14.30 0.4 215 256 M4.5 PX305 99 · · · · · · · · · 305 5:35:07.6 -5:24:01 302 0.45 14.58 0.3 254 137 M3 PV355 0 · · · · · · 403 355 5:35:10.9 -5:22:46 197 0.42 15.03 3.5 294 90 K: X370 99 · · · · · · 452 370 5:35:11.9 -5:19:26 122 0.73 13.86 3.5 344 246 K1.5 PX383 99 · · · V1492 489 383 5:35:12.7 -5:16:14 280 0.19 14.57 2.1 353 433 M3 PXV392 99 · · · · · · 498 392 5:35:12.7 -5:27:11 185 0.23 15.21 0.2 194 234 M6 PX391 99 1806 · · · 501 391 5:35:12.8 -5:20:44 94 0.29 12.97 >3.3 341 168 M1 PXV399 99 · · · · · · 523 399b 5:35:13.2 -5:22:21 345 0.22 16.77 0.1 322 78 · · · PX410 99 · · · · · · · · · 410 5:35:13.2 -5:36:18 167 1.19 16.00 1.3 184 777 · · · PV406 99 · · · V1327 543 406 5:35:13.5 -5:17:10 271 0.95 13.88 2.6 353 375 M1 PXV- - · · · · · · 562 9048 5:35:13.6 -5:21:21 62 0.24 16.27 1.5 341 129 M3 X

422 99 · · · V1495 566 422 5:35:13.7 -5:28:46 74 0.31 14.43 0.0 187 326 · · · PXV436 0 · · · · · · 620 436a 5:35:14.3 -5:22:04 238 0.32 16.97 1.0 338 85 · · · X439 99 · · · V1329 626 439 5:35:14.5 -5:17:25 150 0.30 14.97 0.1 355 359 M1 PXV439 99 · · · V1329 626 439 5:35:14.5 -5:17:25 78 1.21 14.97 4.1 355 359 M1 PXV444 99 · · · · · · 651 444 5:35:14.6 -5:16:46 190 0.18 15.53 1.0 356 398 · · · PX445 26 · · · · · · 645 445a 5:35:14.7 -5:20:42 191 0.44 14.67 1.8 351 163 · · · X- - · · · V1500 · · · 466 5:35:14.9 -5:36:39 82 0.38 14.45 2.2 182 797 · · · V

465 97 1875 V409 · · · 465 5:35:14.9 -5:38:06 271 0.43 14.71 4.1 182 883 · · · PV498 99 1873 V1504 · · · 498 5:35:15.8 -5:32:59 122 0.70 13.81 0.2 181 576 · · · PHV509 99 · · · · · · 789 509a 5:35:16.2 -5:24:56 46 0.49 15.06 0.1 182 93 · · · PXV a,b,f511 99 · · · · · · · · · 511a 5:35:16.3 -5:22:10 241 0.41 15.67 2.4 358 73 M1 PX a,c,e,f- - · · · · · · 822 3031 5:35:16.8 -5:17:17 84 0.26 16.37 0.5 1 366 · · · X

551 99 1914 · · · 881 551a 5:35:17.4 -5:25:45 265 0.14 13.87 0.0 174 142 M1 PX551 99 1914 · · · 881 551a 5:35:17.4 -5:25:45 262 1.27 13.87 3.6 174 143 M1 PX552 99 1908 V410 897 552a 5:35:17.5 -5:21:46 156 0.46 14.50 >1.8 9 99 · · · PXV560 98 · · · V1334 927 560 5:35:17.9 -5:15:33 235 0.60 14.17 3.2 3 471 · · · PXV570 99 · · · · · · 937 570a 5:35:17.9 -5:25:34 0 0.15 14.73 0.8 170 133 · · · PX566 0 · · · · · · 939 566 5:35:18.0 -5:16:13 214 0.86 14.93 0.0 3 430 · · · X- - · · · · · · 967 9224 5:35:18.4 -5:24:27 78 0.42 15.16 2.4 155 70 M2.5 X a,b,f

592 99 · · · · · · 974 592 5:35:18.5 -5:18:21 285 0.61 14.37 4.0 6 304 M2.5 PX597 0 · · · · · · 994 597 5:35:18.8 -5:14:46 307 0.17 14.02 0.1 4 519 · · · XHV- - · · · · · · 998 9239 5:35:18.8 -5:22:23 314 0.20 17.54 0.3 31 70 · · · X- - · · · · · · 1061 5066 5:35:20.0 -5:18:47 71 0.22 17.31 0.1 11 281 M9 0X00

638 99 · · · · · · 1077 638 5:35:20.0 -5:29:12 354 1.01 17.03 3.2 171 353 · · · PXV

15

Table 1—Continued

JW % Par GCVS COUP H97 RA2000 DEC2000 PA SEP [′′] mI ∆mHα1 PAOriC OriC SpT2 Memb.3 Bin

681 99 · · · V1524 · · · 681a 5:35:21.4 -5:23:45 232 1.09 16.39 0.9 107 77 K7 PX0V687 54 · · · · · · 1158 687a 5:35:21.7 -5:21:47 231 0.49 14.66 2.1 39 124 · · · XV a,c709 99 1994 · · · 1202 709 5:35:22.2 -5:26:37 309 0.22 12.92 0.5 156 213 M0.5 PX722 98 · · · · · · 1208 722 5:35:22.3 -5:33:56 216 0.36 14.51 0.2 172 639 M4.5 PX727 99 · · · V1528 1233 727 5:35:22.8 -5:31:37 285 0.73 13.87 4.7 169 503 M2 PXHV748 99 · · · · · · 1279 748a 5:35:24.1 -5:21:33 277 0.35 14.77 2.0 46 159 M3 PXV767 99 · · · · · · 1316 767 5:35:25.2 -5:15:36 75 1.13 13.89 >4.1 16 485 M2.5 PXV776 99 · · · V496 1328 776a 5:35:25.4 -5:21:52 73 0.50 14.20 1.8 56 162 · · · PXV777 80 · · · · · · 1327 777 5:35:25.5 -5:21:36 347 1.19 15.15 1.2 52 173 K6 X783 95 · · · · · · · · · 783 5:35:25.5 -5:34:03 283 0.40 14.17 0.8 168 656 · · · PX797 99 · · · · · · 1363 797 5:35:26.6 -5:17:53 323 1.42 16.30 2.8 25 363 · · · PX- - · · · · · · 1425 3042 5:35:29.5 -5:18:46 329 0.23 16.06 2.6 35 338 · · · XV

841 99 · · · · · · · · · 841 5:35:30.0 -5:12:28 16 0.41 14.41 0.1 17 686 M4 PHV867 99 2075 · · · 1463 867 5:35:31.3 -5:18:56 212 0.29 12.19 >0.3 40 347 K8 PXHV884 99 · · · · · · · · · 884 5:35:32.4 -5:14:25 287 1.32 14.91 2.8 24 589 · · · P893 99 · · · · · · · · · 893 5:35:33.2 -5:14:11 208 0.15 14.07 0.0 24 606 · · · PV906 99 · · · · · · · · · 906 5:35:34.7 -5:34:38 293 1.39 14.17 2.7 158 728 M3 PV924 99 · · · · · · · · · 924 5:35:36.5 -5:34:19 73 1.04 16.63 1.3 156 722 · · · P945 99 2149 · · · · · · 945 5:35:40.2 -5:17:29 46 1.41 12.51 3.6 45 501 B6 P

a - Prosser et al. (1994)b - Padgett et al. (1997)c - Simon et al. (1999)d - Koehler et al. (2006)e - Getman et al. (2005)f - Lucas et al. (2005)

1The symbol > indicates that the primary is saturated (sometimes both stars), thus ∆m is only an approximation

2Spectral types obtained at SIMBAD

3Membership criteria used: P - proper motion catalog by Jones & Walker (1988) with probability ≥ 93%, X - COUP source by Getman etal. (2005), H - Hα emission (Pettersson et al., in prep.), V - variability noted by the General Catalog of Variable Stars

Table 2

Other binaries with angular separations <1.′′5 toward the ONC.

JW % Par GCVS COUP H97 RA2000 DEC2000 PA SEP [′′] mI ∆mHα PAOriC OriC SpT∗ Memb.

553 99 1911 V1510 899 553a 5:35:17.6 -5:22:57 113 0.36 12.41 1.4 34 31 K3.5 PXV596 99 1927 AF 986 596 5:35:18.7 -5:23:14 78 0.30 13.06 2.2 75 34 K3.5 PXV· · · · · · · · · · · · 1085 3075 5:35:20.2 -5:23:09 104 0.21 · · · 0.4 76 57 · · · X182 48 · · · · · · 127 182 5:34:58.0 -5:29:41 77 0.11 16.56 0.2 216 467 · · · X290 99 · · · · · · 266 290 5:35:06.4 -5:27:05 84 0.11 15.48 1.3 214 268 · · · PX- - · · · · · · 1195 3034 5:35:22.3 -5:18:09 227 0.11 15.67 0.8 15 326 · · · X

· · · · · · 1423 · · · · · · 3114 5:34:19.5 -5:27:12 168 0.56 10.52 0.1 255 881 G558 0 · · · · · · · · · 58 5:34:41.8 -5:34:30 282 1.43 16.13 2.2 218 844 · · ·

61 46 · · · · · · · · · 61 5:34:42.7 -5:28:37 153 0.45 13.52 0.9 238 594 · · ·

· · · · · · · · · · · · · · · · · · 5:34:45.8 -5:30:58 324 0.46 · · · 0.1 225 645 · · ·

· · · · · · · · · · · · · · · 10343 5:35:01.4 -5:24:13 292 0.28 · · · 0.6 257 231 · · ·

· · · · · · · · · · · · · · · 9221 5:35:18.3 -5:24:39 96 0.67 16.72 2.2 160 81 · · ·

· · · · · · · · · · · · · · · · · · 5:35:30.0 -5:34:31 237 1.29 · · · 0.3 163 699 · · ·

a - Prosser et al. (1994)1 - Binaries with angular separation <0.

′′4 inside the exclusion zone2 - Binaries with angular separation <0.

′′15 outside the exclusion zone3 - Binaries with angular separation <1.

′′5 outside the exclusion zone but with no evidence of ONC membership

∗Spectral types obtained at SIMBAD

16

Fig. 1.— Distribution of HST images across theONC. The 60′′ exclusion zone centered on θ1 Ori Cis marked by a circle, and the boundaries ofthe COUP survey is indicated with dashed lines.Binaries in three different separation ranges aremarked by different symbols (asterisks: 0.′′1 - 0.′′5,diamonds: 0.′′5 - 1.′′0, triangles: 1.′′0 - 1.′′5). Coor-dinates are equinox 2000.

17

Fig. 2.— Two plots showing ∆m vs. binary sepa-ration in arcseconds. The left panel shows the ac-tual binaries observed, while the right panel shows“observations” of actual stellar images to whichartifical companions were added with random sep-arations, position angles, and ∆m. Filled circlesindicate those companions we could detect, andempty circles those we failed to see. The verticaldashed line indicates the 0.′′15 limit we adopted asour completeness limit.

18

Fig. 3.— All binaries identified among the ONCmembers. Each stamp is 2′′ wide, and in eachpanel, north is up and east is to the left.

19

Fig. 3.— continued

20

Fig. 4.— The surface density of stars in M42 as afunction of distance from θ1 Ori C. Measurementshave been done in 30′′ wide annuli on our ACSimages. The vertical dotted line indicates the 60′′

exclusion zone inside which we did not attempt toidentify visual binaries due to the high density ofstars.

21

Fig. 5.— The probability that a star would haveanother star as a line-of-sight association within aseparation of 1.′′5 for all 781 ONC members in ourlist and as a function of distance from θ1 Ori C.White squares are single stars and black circles areconfirmed binaries. The peak around 180′′ is dueto a subclustering of stars.

22

Fig. 6.— Histogram of binary separations as func-tion of angular separation in steps of 0.′′1. Thenumber of binaries in the innermost bin with sep-arations less than 0.′′1 is incomplete. There is adramatic decrease in the number of binaries whenthe separation increases beyond 0.′′5.

23

Fig. 7.— Logarithmic separation distributionfunction of ONC binaries compared to the distri-bution of field binaries from Duquennoy & Mayor(1991) across the separation range from 0.′′15 to1.′′5. The actual data from Duquennoy & Mayorwithin our observed range are marked by twodashed crosses and the Gaussian distribution thatthey fit to their complete data set is indicated bythe dot-dashed curve. The parts of the columnsthat are dotted indicate the 9 binaries that we cal-culate are due to line-of-sight associations. Theaxis on top indicates separations in arcseconds atthe distance of the ONC. A scaling was done tocorrect for the fact that Duquennoy & Mayor hasa smaller sample size (164 vs. 781) and a largerbin width (0.667 vs. 0.2 in log (AU)), hence a fac-tor of (781/164)×(0.2/0.667) = 1.43 was appliedto compare the distributions directly.

24

Fig. 8.— Cumulative distributions of close (0.′′15- 0.′′5) and wide (0.′′5 - 1.′′5) binaries in the ONCas a function of distance from θ1 Ori C.

25

Fig. 9.— The ratio of wide (0.′′5 - 1.′′5) to close(0.′′15 - 0.′′5) binaries in the ONC as a function ofdistance to θ1 Ori C. The figure shows this ra-tio for all binaries from the 60′′ exclusion zoneand out to a given distance. The last value ataround 900′′ thus represents the value of all bina-ries within the above ranges in the ONC outsidethe exclusion zone. The dashed lines indicate theerrors on the numbers. The dotted horizontal linesrepresent the same ratio from the Duquennoy &Mayor (1991) field binary study, with the lowerbeing from the Gaussian fit, and the upper fromtheir actual data points. The upper scale indicatesthe crossing time in years, assuming a mean one-dimensional velocity dispersion of 2 km/sec. Thevertical line indicates the distance where the ratiobecomes flat, suggesting an age for the ONC ofabout 106 yr.

26

Fig. 10.— Luminosity ratio of the binary popula-tion, based on the Hα fluxes of primaries and sec-ondaries. All stars with separations between 0.′′1and 1.′′5 from Tables 1 and 2 are plotted, except ifthey are saturated.

27